What intelligence leaves out

A brilliant answer arrives at a desk already supported by a world. Someone supplied the question, someone maintains the system, and someone will have to act on the result. Its usefulness depends on those arrangements. When we call the answer intelligent, we can easily stop looking at the conditions that allow intelligence to become useful work.

Imagine a design team using AI to explore possibilities for a small public installation. This is a hypothetical example. The range of ideas helps people see shapes and uses they would otherwise have missed. A proposal wins attention because it combines an elegant form with an inviting account of public benefit. Then someone asks who will maintain it, how it will be repaired, and what happens when its original purpose changes. Those questions concern the future life of the idea.

Benjamin Bratton's Planetary Sapience asks readers to see computation within the material planet. In the Synthetic Intelligence section, he traces its infrastructure through mines, data centers, cables, and interfaces, then calls for attention to physical costs and worthwhile purposes. The argument gives technological possibility a demanding setting: intelligence operates through a world it also helps alter. It does not follow that every use is wasteful or that computation should be refused. The purposes and conditions deserve examination together.

At the scale of a team, this suggests a practical distinction. An answer can be impressive as an answer while asking too much of the setting in which it will be used. A generated plan may assume attention that nobody has available, maintenance that nobody has accepted, or knowledge that the organization has never gathered. These are questions about the proposal's fitness for an actual shared life.

Wisdom begins to matter where knowing more meets choosing what deserves care. A system may help reveal a dependence, compare an alternative, or uncover an error. That contribution can be considerable. The leader still needs to ask which burden is justified by the purpose, who has agreed to carry it, and how that agreement will be revisited. Capability alone cannot decide those questions for the people involved.

Return to the imagined installation. The valuable next step may preserve the ambition while changing its form. A design that local people can repair could sustain participation for longer. A simpler proposal could leave resources for activities that make the place useful. Another proposal might justify greater demands because its benefit is unusually important. The exercise should leave these possibilities open long enough for judgment.

Choose one AI-assisted proposal that your team is considering. Spend twenty minutes writing a short account of its conditions. Begin with the human purpose in a sentence that someone outside the project could understand. Name the people and material arrangements on which it relies. Identify one ongoing burden that the presentation has left vague, and the person who can help you understand it. Mark what you know and what remains unknown.

Finish with a concrete choice: one demand you could reduce, one benefit worth protecting, and one condition that would make you revise the proposal. The output is a one-page account of purpose and upkeep, attached to the decision. Invite someone responsible for the ongoing work to challenge it. Also identify a person outside the project who may bear a burden without sharing its main benefit. Ask them, or an appropriate representative, what would need to change for the arrangement to serve them fairly. Let that answer affect the proposal while change is still possible.

At the next review, look for a change in the proposal or in the reasons for keeping it. Perhaps a maintenance obligation now has an owner. Perhaps the team chooses a more demanding design after explaining why the added burden serves people. Either can represent progress. The test is whether an important condition has entered the decision while there is still time to respond.

This page cannot replace a technical, environmental, safety, or other specialist assessment. Nor should unknown costs be filled with convenient estimates. The first useful result may be a better question for someone qualified to answer it. Keep the inquiry proportional to the decision, and avoid making a small experiment carry the burden of resolving every planetary consequence of computation.

There is a hopeful ambition here. Intelligence can help us perceive more of the world that sustains us: connections across distance, overlooked needs, and consequences that unaided attention might miss. Its promise becomes more interesting when it enlarges our capacity to care for what we discover. The proposal on the desk can then become a beginning for a more considered relationship with the people and conditions that will give it life.